The HaarWavelet Transform in Digital Image Processing:

Its Status and Achievements

Piotr Porwik, Agnieszka Lisowska
Institute of Informatics, University of Silesia, ul. Bedzi nska 39, 41-200 Sosnowiec, Poland
e-mail: porwik@us.edu.pl
Institute of Mathematics, University of Silesia, ul. Bankowa 14, 40-007 Katowice, Poland
e-mail: alisow@ux2.math.us.edu.pl
Abstract. Image processing and analysis based on the continuous or discrete image transforms
are classic techniques. The image transforms are widely used in image ltering, data description, etc.
Nowadays the wavelet theorems make up very popular methods of image processing, denoising and
compression. Considering that the Haar functions are the simplest wavelets, these forms are used in
many methods of discrete image transforms and processing. The image transform theory is a well
known area characterized by a precise mathematical background, but in many cases some transforms
have particular properties which are not still investigated. This paper for the rst time presents graphic
dependences between parts of Haar and wavelets spectra. It also presents a method of image analysis
by means of the waveletsHaar spectrum. Some properties of the Haar and wavelets spectrum were
investigated. The extraction of image features immediately from spectral coecients distribution were
shown. In this paper it is presented that twodimensional both, the Haar and wavelets functions
products man be treated as extractors of particular image features. Furthermore, it is also shown that
some coecients from both spectra are proportional, which simplify slightly computations and analyses.
Key words: Wavelets, Haar Transform, multiresolution.
1. Introduction
The computer and videomedia applications have developed rapidly the eld of multi-
media, which requires the high performance, speedy digital video and audio capabilities.
Nowadays, the image processing and analysis based on continuous or discrete trans-
forms are the classic processing techniques [3, 27, 30, 36, 50]. Digital signal processing
is widely used in many areas of electronics, communication and information techniques
[1, 4, 6, 9, 14, 15, 17, 18, 20, 23, 28, 35]. In the signals compression, digital ltration,
systems identication, the commonly used transforms are based on sinusoidal basic func-
tions such as: Discrete Fourier Transform, Discrete Sine or Cosine Transform, Hartley
Transform or rectangular basic functions Slant Transform, Discrete Walsh Transform,
a nd Discr ete Wavelet Tr ansfor m (Haar , Daub echies, etc.) [3, 4, 6, 11, 12, 17, 25, 49, 50].
All these mentioned functions are orthogonal, and their forward and inverse transforms
require only additions and subtractions. It makes that it is easy to implement them on
the computer.
M ach in e GRAP HI CS & VI S I O N vol . 13, n o. 1/ 2, 2004, p p . 79- 98
80 T he HaarWavelet Transform in Digital Im age Processing: It s Statu s and Achievem ents
Haar functions are used since 1910. They were introduced by Hungarian mathemati-
cian Alfred Haar [1]. Nowadays, several denitions of the Haar functions and various
generalizations [39] as well as some modications [19, 37, 50] were published and used.
One of the best modication, which was introduced, is the lifting scheme [25, 26, 29].
These transforms have been applied, for instance, to spectral techniques for multiple
valued logic, image coding, edge extraction, etc. Over the past few years, a variety of
powerful and sophisticated waveletbased schemes for image compression, as discussed
later, were developed and implemented. Wavelet scheme gives many advantages, which
are used in the JPEG2000 standard as waveletbased compression algorithms [31].
Generally, wavelets, with all generalizations and modications, were intended to
adapt this concept to some practical applications [40, 42]. The Discrete Wavelet Trans-
form uses the Haar functions in image coding, edge extraction and binary logic design
and is one of the most promising technique today. The nonsinusoidal Haar transform
is the complete unitary transform [15, 16, 17]. It is local, thus can be used for data
compression of nonstationary spiky signals. The digital images may be treated as
such spiky signals. Unfortunately, the Haar Transform has poor energy compaction
for image, therefore in practice, basic Haar transform is not used in image compression.
One should remember that researches in this topic are still in progress and everyday new
solutions and improvements are found [33, 39, 41, 43, 47].
Fourier methods are not always good tools to recapture the signal [3], particularly if it
is highly nonsmooth; too much Fourier information is needed to reconstruct the signal
locally. In these cases the wavelet analysis is often very eective because it provides
a simple approach for dealing with the local aspects of a signal, therefore particular
properties of the Haar or wavelet transforms allow to analyse the original image on
spectral domain eectively. These methods will be described in this paper.
2. The Discrete Haar Transform
A complete orthogonal system of functions in L
p
[0, 1], p [0, ] which take values from
the set {0, 2
j
: j N} was dened by Haar [1]. This system of functions has property
that each function continuous on interval [0, 1] may be represented by a uniformly and
convergent series in terms of elements of this system. Nowadays, in the literature, there
are some other denitions of the Haar functions [16]. Those denitions are mutually
diering with respect to the values of Haar functions at the points of discontinuity. For
example the original Haar denition is as follows [4]:
haar(0, t) = 1, for t [0, 1); haar(1, t) =
_
1, for t [0,
1
2
),
1, for t [
1
2
, 1)
(1)
and haar(k, 0) = lim
t0+
haar(k, t), haar(k, 1) = lim
t1
haar(k, t) and at the points
of discontinuity within the interior (0, 1) haar(k, t) =
1
2
(haar(k, t 0) +haar(k, t +0)).
M ach i n e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p .79-98
Piotr Porwik, Agnieszka L isowska 81
Instead of described relations some authors use the formula haar(k, t) = haar(k, t + 0)
where in the practice it is usually assumed that the Haar function takes zero value at
the points of discontinuity.
Discrete Haar functions may be dened as functions determined by sampling the
Haar functions at 2
n
points. These functions can be conveniently represented by means
of matrix form. The Haar matrices H(n) are considered in the natural and sequence
ordering which dier in the ordering of rows. Each row of the matrix H(n) includes
the discrete Haar sequence haar(w, t) (or otherwise the discrete Haar function). In this
notation, index w identies the number of the Haar function and index t the discrete point
of the function determination interval. In this case, the Haar matrix of any dimension
may be obtained by the following recurrence relation [3]:
H(n) =
_
H(n 1) [1 1]
2
(n1)/2
I(n 1) [1 1]
_
, H(0) = 1 (2)
and H(n) = H(n)
T
for n > 1 and H(n)
1
= 2
n
H(n)
T
, where H(n) matrix of
the discrete Haar functions of degree 2
n
, I(n) identity matrix of degree 2
n
, the
Kronecker (tensor) product.
From (2) one can observe that, unlike the Fourier transform, H(n) matrix has only
real elements. The Haar matrix is nonsymmetric and its elements are 1, 1 or 0,
multiplied by powers of

2. The discrete, orthogonal Haar functions, obtained from the

formula (2), are dened on [0, 1) interval.
Def. 2.1. Twodimensional N N = 2
n
2
n
forward discrete Haar transform is
dened in matrix notation as
S = a H(n) F a H(n)
T
. (3)
The inverse transform is dened as
F = b H(n)
T
S b H(n) , (4)
where F is the image in matrix form, the matrix is of dimension N N pixels, S is the
spectrum matrix and a b = 1/N, hence parameters a or b may be dened as values 1/N,
1/

N or 1, n = log
2
N.
Because matrix (2) has many zero entries, some values of the Haar spectral coecients
are equal to 0 too. Fig. 1 presents some known transforms of a test image. As transform
matrices H(n), WalshHadamard, Haar, sine and cosine matrices were used. All gures
make up the graphic representation of spectral coecients. Therefore each picture in
Fig. 1 may be interpreted as matrix, where axis row and col describe elements of matrix
S and axis value of coe. indicates values of spectral coecients in S. The test image F
is built as 8 8 matrix, which has zero values everywhere except the upper left element,
which has the value of eight.
The Fast Haar Transform has already been well known from many works [3, 4, 5,
17, 28] therefore it will not be present in details. The Haar transform is a symmetric,
M ach in e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p .79-98
82 T he HaarWavelet Transform in Digital Im age Processing: It s Statu s and Achievem ents
separable transform that uses Haar function for its basis. It exists for N = 2
n
, where n is
an integer. The Haar function, which is an odd rectangular pulse pair, is the simplest and
oldest orthonormal wavelet [16, 17, 28]. Whereas the Fourier transform basis functions
dier only in frequency, the Haar functions vary in the both scales of width and position.
a) a = 1/

8 b) a = 1/

8
c) a = 1
d) a = 1
Fig. 1. The S transform of image containing the test impulse: a) WalshHadamard, b) Haar, c) DST
(Discrete Sine Transform), d) DCT (Discrete Cosine Transform).
From Fig. 1 we can observe that all N
2
elements of these transforms are nonzero
except the Haar transform, which has only 2N nonzero entries. These features are very
important in image processing because in many cases spectral coecients have zero
entries before next retrieval operations. This case occurs in black and white images very
often. From image data compression point of view, this case is very convenient. From
the energy distribution it can be estimated which spectral coecients can be reduced
[15, 17, 28]. It is easy to observe from Fig. 1 that the Walsh transform gives the worst
results here: distribution of spectral energy is uniformable. In c) and d) cases distribution
M ach i n e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p .79-98
Piotr Porwik, Agnieszka L isowska 83
of spectral energy has sharply outlined maximum. Outside of the maximum value, one
can observe the decrease of energy. Unfortunately, the distribution of the Haar spectrum
(Fig. 1b) is not proper too, but we can treat this transform dierently.
The discrete transforms, presented in Fig. 1, enable us to observe where energy
concentrations occur. Hence, we know which spectral coecients are important in image
processing. From this representation, it is not possible to nd more precisely information
about real image. For example, it is dicult to point places, which describe horizontal,
vertical, etc. details of real image. These troubles can be overcome by well known
multiresolution analysis [11, 17, 20, 34, 49].
3. The Discrete Haar Wavelet Transform
An outstanding property of the Haar functions is that except function haar(0, t), the
ith Haar function can be generated by the restriction of the (j 1)th function to
be half of the interval where it is dierent from zero, by multiplication with

2 and
scaling over the interval [0, 1]. These properties give considerable interest of the Haar
function, since they closely relate them to the wavelet theory. In this setting, the rst
two Haar functions are called the global functions, while all the others are denoted
as the local functions. Hence, the Haar function, which is an odd rectangular pulse
pair, is the simplest and oldest wavelet. The motivation for using the discrete wavelet
transform is to obtain information that is more discriminating by providing a dierent
resolution at dierent parts of the timefrequency plane. The wavelet transforms allow
the partitioning of the time-frequency domain into nonuniform tiles in connection with
the timespectral contents of the signal. The wavelet methods are strongly connected
with classical basis of the Haar functions; scaling and dilation of a basic wavelet can
generate the basis Haar functions.
Def. 3.1. Let : R R, the Haar wavelet function is dened by the formula [43]:
(t) =
_
_
_
1, for t [0,
1
2
),
1, for t [
1
2
, 1),
0, otherwise.
(5)
Taking into account the Denition 3.1, any Haar function (except function haar(0, t))
from basis (2) may be generated by means of the formulas:

j
i
(t) =

2
j
(2
j
t i), i = 0, 1, . . . , 2
j
1 and j = 0, 1, . . . , log
2
N 1 . (6)
The constant

2
j
is chosen so that the scalar product <
j
i
,
j
i
>= 1,
j
i
L
2
(R). If
one considers the wavelet function on other intervals than [0, 1), the normalisation con-
stant will be dierent. For example:
0
0
= haar(1, t),
1
0
= haar(2, t),
1
1
= haar(3, t),
M ach i n e GRAP HI CS & VI S I O N vol . 13, n o. 1/ 2, 2004, p p . 79- 98
84 T he HaarWavelet Transform in Digital Im age Processing: It s Status and Achievements

2
0
= ha ar(4, t),
2
1
= haar(5, t),
2
2
= haar(6, t),
2
3
= haar(7, t). Generally
j
i
=
haar(2
j
+i, t). From this example follows that functions
j
i
(t) are orthogonal to one an-
other. Hence, we obtain linear span of vector space W
j
= span{
j
i
}
i=0,...,2
j
1
. A collec-
tion of linearly independent functions {
j
i
(t)}
i=0,...,2
j
1
spanning W
j
is called wavelets.
Def. 3.2. Let : R R, the Haar scaling function is dened by the formula [43]:
(t) =
_
1, for t [0, 1),
0, for t [0, 1).
(7)
Similarly to the properties of the wavelet function, for scaling function one can dene
the family of functions:

j
i
(t) =

2
j
(2
j
t i), i = 0, 1, . . . , 2
j
1 and j = 0, 1, . . . , log
2
N 1 . (8)
The constant

2
j
is chosen so that the scalar product <
j
i
,
j
i
>= 1,
j
i
L
2
(R). The
index j refers to dilation and index i refers to translation [11, 17]. Hence, we obtain
linear span of vector space V
j
= span{
j
i
}
i=0,...,2
j
1
. The basic functions from the
space V
j
are called scaling functions.
In multiresolution analysis the Haar basis has important property: V
j
= V
j1

W
j1
, where stands for orthogonality of V
j
and W
j
spaces [16].
From Denitions 3.1 and 3.2 follows, that vector space W
j
can be treated as the or-
thogonal complement of V
j
in V
j+1
. In other words, let W
j
be the space of all functions
in V
j+1
that are orthogonal to all functions in V
j
. Therefore, the basis functions
j
i
of
W
j
together with the basis functions
j
i
of V
j
form a basis for V
j+1
and every basis
function
j
i
of W
j
is orthogonal to every basis function
j
i
of V
j
. From the properties of
the Haar functions, described above, follows that basic wavelet is progressively narrowed
(reduced in scale) by powers of two. Each smaller wavelet is then translated by incre-
ments equal to its width, so that the complete set of wavelets at any scale completely
covers the interval. From mentioned equations one can conclude, that the basic wavelet
is scaled down by powers of 2, but its amplitude is scaled up by powers of

2.
4. The Haar and Wavelet Basic Images
Due to its low computing requirements, the Haar transform has been mainly used for
image processing and pattern recognition. From this reason two dimensional signal
processing is an area of ecient applications of Haar transforms due to their wavelet
like structure.
Because H(n) and H(n)
T
are the square matrices, their product is commutative,
therefore equations (3) and (4) can be rewritten and expressed as:
M ach i n e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p.79-98
Piotr Porwik, Agnieszka L isowska 85
s(k, m) =
N1

x=0
N1

y=0
f(x, y) haar(k, x) haar(m, y) , (9)
f(x, y) =
N1

k=0
N1

m=0
s(k, m) haar(k, x) haar(m, y) , (10)
where S = [s
km
], F = [f
xy
], x, y, k, m {0, 1, . . . , N 1}.
Basing on equation of analysis (3) we can conclude that in 2D spectral domain the
values of coecients depend on appropriate product of the two Haar functions. Fig. 2
presents an example of multiplication the arbitrary selected Haar functions.
haar(4,x)
x
y
h
a
a
r
(
1
,
y
)
<0
>0
=0
Fig. 2. The example of multiplication of the discrete Haar functions.
Because this product is multiplied by image matrix, the result of such multiplication
may be treated as a particular extractor, which allows us to characterise the features
in real image at specied point. In other words, that product can be used to locate
the specic lines (edges) hidden in image. By looking for all coecients in the spectral
space, one can nd all important edge directions in the image. This information can
further be used in image (object) analysis [10].
Taking into account the Mallat algorithm [7, 8, 9] multiresolution image analysis
can be applied to the classical Haar basis, described by matrixequation (2). In this
case, we must nd decomposition matrices of matrix . Based on Fig. 2, construction of
decomposition matrices can be as follows:
Step 1. According to the formula V
n
= V
n1
W
n1
, the matrix M
1
has a form
M
1
= [
n1
j=0,...,2
n1
1
V
n1
,
n1
j=0,...,2
n1
1
W
n1
]
T
.
M ach i n e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p .79-98
86 T he HaarWavelet Transform in Digital Im age Processing: It s Statu s and Achievem ents
Step 2. In second step, because V
n1
= V
n2
W
n2
W
n1
, the matrix M
2
can
be constructed as follows
M
2
= [
n2
j=0,...,2
n2
1
V
n2
,
n2
j=0,...,2
n2
1
W
n2
,
n1
j=0,...,2
n1
1
W
n1
]
T
.
. . .
Step n. Finally, after n steps of calculations V
1
= V
0
W
0
W
1
W
2
. . . W
n1
,
hence the matrix M
n
has a structure
M
n
= [
0
0
V
0
,
0
0
W
0
,
1
j=0,1
W
1
, . . . ,
n1
j=0,...,2
n1
1
W
n1
]
T
.
Exmp. 4.1. Let n = 3 then:
V
3
= V
2
W
2
M
1
= [
2
0
,
2
1
,
2
2
,
2
3
,
2
0
,
2
1
,
2
2
,
2
3
]
T
=
_

_
2 2 0 0 0 0 0 0
0 0 2 2 0 0 0 0
0 0 0 0 2 2 0 0
0 0 0 0 0 0 2 2
2 2 0 0 0 0 0 0
0 0 2 2 0 0 0 0
0 0 0 0 2 2 0 0
0 0 0 0 0 0 2 2
_

_
V
2
= V
1
W
1
W
2
M
2
= [
1
0
,
1
1
,
1
0
,
1
1
,
2
j=0,...,3
W
2
]
T
=
_

2 0 0 0 0
0 0 0 0

2 0 0 0 0
0 0 0 0

2
2 2 0 0 0 0 0 0
0 0 2 2 0 0 0 0
0 0 0 0 2 2 0 0
0 0 0 0 0 0 2 2
_

_
V
1
= V
0
W
0
W
1
W
2
M
3
= [
0
0
,
0
0
,
1
j=0,1
W
1
,
2
j=0,...,3
W
2
]
T
=
M ach i n e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p .79-98
Piotr Porwik, Agnieszka L isowska 87
_

_
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1

2 0 0 0 0
0 0 0 0

2
2 2 0 0 0 0 0 0
0 0 2 2 0 0 0 0
0 0 0 0 2 2 0 0
0 0 0 0 0 0 2 2
_

_
For last decomposition level, it can be noticed that M
n
= H(n). If each orthogonal ma-
trix M
i
, i = 1, 2, 3 will be multiplied by 1/

2 factor, then the procedure of calculations

will be according to the classical Mallat algorithm.
The product of the decomposition levels for all 2D Haar functions (for case N = 8)
is shown in Fig. 3. Taking into account equation (9), these products can be treated as
extractors of image features.
Fig. 3. The 2D Haar functions product treated as extractors. Decomposition levels: (left) rst, (center)
second, (right) third.
The pictures presented in Fig. 3 have been composed on the basis of M
i
, i = 1, 2, 3
matrices and method shown in Fig. 2.
From Fig. 3 we can conclude that the classical Haar transform gives dierent spectral
coecients on dierent decomposition levels. One advantage of the method presented
above is that often a large number of the detail coecients turn out to be very small in
magnitude, as in the example of Fig. 1. Truncating, or removing, these small coecients
from the representation introduces only small errors in the reconstructed image, giving
a form of lossy image compression. Additionally, we can control which coecients will
be removed, because its distribution is known (Fig. 3). In other words, that distribution
M ach in e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p .79-98
88 T he HaarWavelet Transform in Digital Im age Processing: It s Statu s and Achievem ents
may be used to locate the specic lines (edges) hidden in image. By looking for all large
coecients in the spectral space, it can be found all important edge directions in the
image.
There are many ways of forming the Haar spectrum. The simplest method consists of
multiplying the data by the Haar matrices, then setting some of the resulting coecients
equal to zero (Fig. 1). A widely used method involves the specifying threshold [12, 15, 17].
All coecients whose magnitudes lie bellow this threshold are set equal to 0. This method
is frequently used for noise removal, where coecients whose magnitudes are signicant
only, because of the added noise will often lie bellow a wellchosen threshold [15]. A
second method keeps only the largest magnitude coecients, while setting the rest equal
to zero. This method is convenient for making comparison when it is known in advance
how many terms are needed. A third method is called the energy method [15, 17]. This
one involves specifying a fraction of the signals energy, where the energy is the square
root of the sum of squares of the coecients. Then we retain the least number of the
largest magnitude coecients whose energy exceeds this fraction of the signals energy
and set all other coecients equal to 0. Instead of the abovedescribed methods of
forming the Haar spectrum, we can apply the presented in this paper method of the 2D
Haar function analysis. This new method allows very often obtaining the similar results
through selection of the appropriate coecients.
Problem of image analysis described above can be solved dierently by applying the
same functions and spaces. As it is known, there are two ways of forming the Haar
coecients: standard and nonstandard method [16]. The standard approach has been
presented above. In this paper is applied nonstandard method because, as it will be
shown, it has a strong connection with method of the Haar functions analysis described
above. Additionally, nonstandard method is slightly more ecient to compute spectral
coecients [49].
Basing on Denitions 3.1 and 3.2, on properties of the Haar functions and the facts
that V
j
= span{
j
i
}
i=0,...,2
j
1
, W
j
= span{
j
i
}
i=0,...,2
j
1
we can express functions
and as a linear combination of the basis functions from V and W spaces respectively.
In the Haar wavelet case and functions can be written as
(t) = h(0)

2(2t) + h(1)

2(2t 1) , (11)
(t) = g(0)

2(2t) + g(1)

2(2t 1) , (12)
where {h(0), h(1)} and {g(0), g(1)} dene the lowpass and highpass lters respectively.
For this case, h(k), k = 0, 1 coecients are known [5]:
h(0) =
1

2
, h(1) =
1

2
, g(0) =
1

2
, g(1) =
1

2
. (13)
Let us denote F as an image in matrix form and dene the operators
M ach i n e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p . 79-98
Piotr Porwik, Agnieszka L isowska 89
A(i) =
1

2
F(2i) +
1

2
F(2i + 1) , (14)
D(i) =
1

2
F(2i)
1

2
F(2i + 1) , (15)
where F(i) vector of size N, containing row or column of matrix F, i {0, 1, . . . ,
N
2

1}, A(i) vector of size N/2, containing approximation coecients, D(i) vector of
size N/2, containing detail coecients.
To get nonstandard wavelet decomposition on the rst level of an image F (the
spectrum matrix called S
1
) we rst apply the operators (14), (15) to all columns of the
matrix and then to all rows [16]. To get the second level of nonstandard decomposition
(that is matrix S
2
) one can apply similar analysis to upper left submatrix of size
N
2

N
2
of matrix S
1
. And generally, to get kth level matrix S
k
, one can apply this analysis
to upper left submatrix of size
N
2
k1

N
2
k1
of matrix S
k1
, where k {1, . . . , log
2
N}.
Note, that applying lters (14), (15) to an image, give the same results as multiplying
matrices S
1
=
1
8
M
1
F M
T
1
, where matrix M
1
is taken from Ex. 4.1. Therefore, S
1
may be treated as extractor of image features on the rst level of wavelet decomposition,
similar as above in Haar decomposition case. Because on the second and next levels only
the part of a matrix is transformed (opposite to Haar decomposition) these extractors
on these levels are dierent. For example, for N = 8 the products of the nonstandard
wavelet decomposition levels are shown in Fig. 4.
Fig. 4. The 2D wavelet functions product treated as extractors. Decomposition levels: (left) rst,
(center) second, (right)third.
All considerations, which have been presented until now for the classical of Haar
functions, have applications in that case too, with the exception of extractors distribution
(Fig. 4). Additionally, it is common knowledge that coecients h(k), k = 0, 1 (in other
M ach in e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p .79-98
90 T he HaarWavelet Transform in Digital Im age Processing: It s Statu s and Achievem ents
wo r ds s o ca lled lter s ) may b e des cr ib ed by mea ns o f s ca la r pr o duct o f functio ns a nd
respectively:
h(k) =< (t),

2(2t k) > (16)

g(k) =< (t),

2(2t k) > (17)

Presented both the standard and nonstandard approaches are applied very often in
practical solutions [5, 15, 16, 17, 28, 46]. Each of them has some advantages. The stan-
dard decomposition can be simply implemented as matrix multiplication. This presents
equation (3). The nonstandard algorithm is a bit more ecient. More compact de-
scription of the nonstandard method allows us even increase ecient computation.
5. Experimental Results
In order to test Haar matrixbased method and wavelet approach the well known image
benchmarks were used. Mentioned benchmarks one can nd in many databases. As test
images we used some standard images such as Barbara, Bridge, Lena, Flowers, Mandrill,
Baboon etc. Each of these images was of size aa8 bits, where a {32, 64, 128, 256},
respectively.
By analysing the Figs. 34 we can divide area of each gure into 4 subsampled
spaces. Each piece has dimension (N/2)(N/2) and is called A, H, V and D respectively.
Location of these subareas presents Fig 5.
A H
V D
Fig. 5. Principle of spectra partitioning.
Each mentioned piece (A, H, V or D) for N = 8 includes sixteen (4 4) appropriate
subsquares from Fig. 34. According to presented previously arguments, mentioned ar-
eas possesses dierent features: A (Approximation Area) includes information about
the global properties of analysing image. Removing spectral coecients from this area
leads to the biggest distortion in origin image. H (Horizontal Area) includes informa-
tion about the vertical lines hidden in image. Removing spectral coecients from this
M ach i n e GRAP HI CS & VI S I O N vol. 13, n o. 1/2, 2004, p p .79-98
Piotr Porwik, Agnieszka L isowska 91
area excludes horizontal details from origin image. V (Vertical Area) contains infor-
mation about the horizontal lines hidden in image. Removing spectral coecients from
this area eliminates vertical details from origin image. D (Diagonal Area) embraces
information about the diagonal details hidden in image. Removing spectral coecients
from this area leads to the smallest distortions in origin image.
Fig. 6 presents Baboon the one of 256 256 8 greylevel test images together
with its wavelet and Haar spectra. The spectra images are dierent what directly follows
from Figs. 34. Taking into account mentioned above features of areas some dierences
between spectres can be shown.
Fig. 6. Original image and its wavelet and Haar spectra, respectively.
Each left image from Figs. 710 presents place where spectrum was reduced. In
this place, all spectral coecients equal to zero. Two remained images show dierences
between origin image and the compressed one for wavelet and Haar method of analysis
respectively.
Fig. 7. Horizontal details elimination and lost information after applied wavelet and Haar matrixbased
m etho d, r es p ecti vel y.
M ach in e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p.79-98
92 T he HaarWavelet Transform in Digital Im age Processing: It s Statu s and Achievem ents
Fig. 8. Vertical details elimination and lost information after applied wavelet and Haar matrixbased
method, respectively.
Fig. 9. Diagonal details elimination and lost information after applied wavelet and Haar matrixbased
method, respectively.
Fig. 10. Horizontal and vertical details elimination and lost information after applied wavelet and Haar
matrixbased method, respectively.
M ach in e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p .79-98
Piotr Porwik, Agnieszka L isowska 93
The other cases of removing the spectral coecients can be applied as well. These
entire processes base on the fact, that appropriate selection and modication of the
spectral coecients may preserve the contents of the image. Depending on our needs
we can shape compression ratio by means of natural selection of spectral coecients
directly from areas A, H, V, D or from their combinations. The exact information about
distribution of spectral coecients allows us to match easily up the compression ratio
with the type of image, but this problem was not considered in the paper. In particular
cases it is possible to obtain the large image compression ratio. Obtained results for
Haar matrixbased method and wavelet method were compared by means of PSNR
coecients (hence there are some other similarity coecients, for example like in [45]).
These results of investigations are collected in Table 1. From this table one can see that
Haar reconstructed images have slightly better quality.
Horizontal Vertical Diagonal Horiz.+Vert.
Method (H) (V) (D) (H+V)
Wavelet decomp. 29, 7254 27, 3697 31, 4822 25, 3813
Haar decomp. 29, 7269 27, 3702 31, 4827 25, 3813
Tab. 1. The PSNR of reconstructed images after appropriate details elimination.
Between Haar matrixbased method and wavelet approach the quantitative and
graphic relationship may be observed. Fig. 11 presents those relations. The image from
Fig. 11 was constructed as a quotient between wavelet and Haar spectra respectively.
These relationships will be obviously identical for any image objects.
Fig. 11. Areas of the proportional coecients (white squares) for Haar and wavelet spectres.
Moreover, it can be proven that for all diagonal coecients on all levels of decompo-
sition the following proposition is true.
M ach in e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p .79-98
94 T he HaarWavelet Transform in Digital Im age Processing: It s Statu s and Achievem ents
Propos. 5.1. Let d
H
stands for diagonal coecients from Haar spectrum matrix
and d
W
stands for diagonal coecients from wavelet spectrum matrix, both of degree
2
n
. Then
d
H
= 2
n
d
W
. (18)
The proof of this fact, describing graphic dependences, follows instantly from Figs. 3
4 and arguments mentioned above, as well as appropriate proportionality of submatrices
from the Haar and wavelet matrices on the same levels of decomposition.
This relationship is well visible in Fig. 11. The white areas appear just due to
the proportionality of diagonal coecients. Furthermore, note from Table 1 (the last
column), that after removing all horizontal and vertical details on the rst level of
decomposition we get exactly the same PSNR of both methods reconstructed images
because of the proportionality of the diagonal detail coecients.
The amount of proportional coecients can be easily computed. It is easy to see
that the number of diagonal detail coecients in spectrum matrix of size 2
n
2
n
equals

n1
i=0
2
i
2
i
. Hence, after simple calculations we get the following corollary.
Corollary 5.1. The number of proportional coecients in Haar and wavelet matri-
ces of size 2
n
2
n
equals

n1
i=0
2
i
2
i
2
n
2
n

1
3
. (19)
It means that as much as 1/3 of all Haar matrix coecients are proportional to wavelet
matrix coecients; for all of them the formula (18) is true.
6. Concluding Remarks
Wavelets analysis is a particular timespace scale representation of signals, which has
found a wide range of applications in physics, signal processing and applied mathematics
in the last years. The purpose of this paper was to compare the wavelets and the Haar
functions in twodimensional space. In many cases, these methods are similar. Presented
in this paper the Haar matrixbased method and wavelet analysis can be used in many
areas of image processing such as denoising, edge detection, edge preserving smoothing
or ltering. In the paper is shown the graphic way of presentation of decomposition levels
for both the Haar matrixbased method and wavelets. As it was shown both methods
may be modelled on the basis of the wavelets theorem.
The 2D Haar matrix method of calculations like the 2D Fast Fourier Transform has
complexity O(4N
2
log
2
N) [14], classical 2 1D fast nonstandard wavelet method of
calculations has complexity O(16/3N
2
) only [28]. This complexity can be decreased to
O(14/3N
2
) by suitable organization of calculations [49]. Described complexity factors
are determined as number of additions and multiplications in computation process. The
graphic distribution of the Haarwavelet spectral coecients also was presented. Ad-
ditionally, graphic presentation of spectra distribution allows us to point appropriate
M ach i n e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p.79-98
Piotr Porwik, Agnieszka L isowska 95
selection or modication (reduction) of the Haarwavelet coecients. For example it
may be removed vertical, horizontal or diagonal details of a given image.
The authors believe that this survey and problem description can be useful to re-
searchers working in dierent disciplines where the discrete Haar and wavelets transforms
are used.
References
1910
[1] Haar A.: Zur Theorie der orthogonalen Funktionensysteme. Mathematische Annalen, 69, 331371.
1973
[2] Rudin W.: Functional Analysis. McGrawHill, New York.
1975
[3] Ahmed N., Rao K.R.: Orthogonal Transforms for Digital Signal Processing. SpringerVerlag, Berlin.
1977
[4] Harmuth H.F.: Sequency Theory. Foundations and applications. Academic Press, New York.
1978
[5] Pratt W. K.: Digital Image Processing. John Wiley and Sons, New York.
1988
[6] Besslich P. W., Trachtenberg L. A.: The sign transform an invertible nonlinear transform with quan-
tized coecients. In: Moraga C., editor, Theory and application of spectral techniques, University
Dortmund Press.
[7] Jain A.K.: Fundamentals of Digital Image Processing. Prentice Hall.
1989
[8] Mallat S.: Multifrequency channel decompositions of images and wavelet models. IEEE Transaction
in Acoustic Speech and Signal Processing, 37, 20912110.
[9] Mallat S. A.: Theory for Multiresolution Signal Decomposition: The Wavelet Representation. IEEE
Transactions on Pattern Analysis and Machine Intelligence, 11(7), 674693.
[10] Yankowitz D., Bruckstein A.M.: A New Method for Image Segmentation. Computer Vision, Graphics
and Image Processing, 46/1, 8295.
1992
[11] Daubechies I.: Ten lectures on wavelets. Philadelphia PA, SIAM.
[12] Grochenig K., Madych W. R.: Multiresolution Analysis, Haar bases and selfsimilar tilings of R
n
.
IEEE Transactions on Information Theory, 38(2), 556568.
1994
[13] Odegard J.: Image Enhancement by Nonlinear Wavelet Processing. Rice University CML Technical
Report.
1995
[14] Bhaskaran V., Konstantinides K.: Image and video compression standards: algorithms and archi-
tectures. Kluwer, Boston.
[15] Stollniz E. J., DeRose T. D., Salesin D. H.: Wavelets for Computer Graphics: A Primer, Part 1.
IEEE Computer Graphics and Applications, 7684.
[16] Fournier A.: Wavelets and their Applications in Computer Graphics. SIGGRAPH95 Course Notes,
University of British Columbia.
1996
[17] Castleman K. R.: Digital Image Processing. PrenticeHall, New Jersey.
1997
[18] Calderbank A. R., Daubechies I., Sweldens W., Yeo B. L.: Lossless image compression using integer
to integer wavelet transforms. Proceedings of the International Conference on Image Processing, 1,
596599.
[19] Claypoole R., Davis G., Sweldens W., Baraniuk R.: Adaptive Wavelet Transforms for Image Coding.
Asilomar Conference on Signals, Systems and Computers.
M ach in e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p .79-98
96 T he HaarWavelet Transform in Digital Im age Processing: It s Statu s and Achievem ents
[20] Creusere C. D.: A new method of robust image compression based on the embedded zerotree wavelet
algorithm. IEEE Transactions on Image Processing, 6, 14361442.
[21] Przelaskowski A., Kazubek M., Jamrogiewicz T.: Optimalization of the WaveletBased Algorithm
for Increasing the Medical Image Compression Eciency. Proceedings of the TFTS97 2nd IEEE
UK Symposium on Applications of TimeFrequency and TimeScale Methods, 177-180.
[22] Walker J. S.: Fourier Analysis and Wavelet Analysis. Notices of the American Mathematical Society,
44(6), 658670.
[23] Wojtaszczyk P.: A Mathematical Introduction to Wavelets. Cambridge University Press, Cambridge.
1998
[24] Calderbank A. R., Daubechies I., Sweldens W., Yeo B. L.: Wavelet transforms that map integers to
integers. Applied and Computational Harmonics Analysis, 5(3), 332369.
[25] Daubechies I.: Recent results in wavelet applications. Journal of Electronic Imaging, 7(4), 719724.
[26] Daubechies L., Sweldens W.: Factoring wavelet transforms into lifting steps. J. Fourier Anal. Appl.,
4(3), 247269.
[27] Davis G., Nosratinia A.: WaveletBased Image Coding: An Overview. Applied and Computational
Control, Signals, and Circuits, 1(1).
[28] Sonka M., Hlavac V., Boyle R.: Image processing, Analysis and Machine Vision. Brooks/Cole Pub-
lishing Comp.
1999
[29] Davis G., Strela V., Turcajova R.; Multivwavelet Construction via the Lifting Scheme. Wavelet
Analysis and Multiresolution Methods, T. X. He (editor), Lecture Notes in Pure and Applied Math-
ematics, Marcel Dekker.
[30] Smith S. W.: The Scientist and Engineers Guide to Digital Signal Processing. California Technical
Publishing, San Diego.
2000
[31] Christopoulos C., Skodras A., Ebrahimi T.: The JPEG2000 Still Image Coding System: an Overview.
IEEE Transactions on Consumer Electronics, 46(4), 11031127.
[32] Fernandes F., van Spaendonck R., Burrus C.: Directional ComplexWavelet Processing. Wavelet
Applications in Signal and Image Processing.
[33] Jahromi O. S., Francis B. A., Kwong R. H.: Algebraic theory of optimal lterbanks. Proceedings of
IEEE International Conference on Acoustics, Speech and Signal Processing, 1, 113116.
[34] Romberg J., Choi H., Baraniuk R., Kingsbury N.: Multiscale Classication using Complex Wavelets.
IEEE International Conference on Image Processing, Vancouver, Canada.
2002
[35] Addison P. S., Watson J. N., Feng T.: LowOscillation Complex Wavelets. Journal of Sound and
Vibration, 254(4), 733762.
[36] Blu T., Unser M.: Wavelets, Fractals and Radial Basis Functions, IEEE Transactions on Signal
Processing, 50(3), 543553.
[37] Munoz A., Ertle R., Unser M.: Continuous wavelet transform with arbitrary scales and O(N) com-
plexity. Signal Processing, 82, 749757.
[38] Resniko H., Wells R. O. Jr.: Wavelet Analysis. SpringerVerlag, New York.
[39] Zeng L., Jansen C. P., Marsch S., Unser M., Hunziker R.: FourDimensional Wavelet Compression of
Arbitrarily Sized Echocardiographic Data. IEEE Transactions on Medical Imaging, 21(9), 11791188.
2003
[40] Drori I., Lischinski D.: Fast Multiresolution Image Operations in the Wavelet Domain. IEEE Trans-
actions on Visualization and Computer Graphics, 9(3), 395411.
[41] Fernandes F., Selesnick I., van Spaendonck R., Burrus C.: Complex Wavelet Transforms with Allpass
Filters. Signal Processing, 83(8), 16891706.
[42] Jorgensen P.: Matrix Factorizations, Algorithms, Wavelets. Notices of the American Mathematical
Society, 50(8), 880894.
[43] Jahromi O. S., Francis B. A., Kwong R. H.: Algebraic theory of optimal lterbanks. IEEE Transac-
tions on Signal Processing, 51, 442457.
[44] Lisowska A.: Nonlinear Weighted Median Filters in Dyadic Decomposition of Images. Annales UMCS
Informatica AI, 1, 157164.
[45] Prieto M. S., Allen A. R.: A Similarity Metric for Edge Images. IEEE Transactions on Pattern
Analysis and Machine Intelligence, 25(10), 12651273.
M ach in e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p .79-98
Piotr Porwik, Agnieszka L isowska 97
[46] Romberg J., Wakin M., Baraniuk R.: Multiscale Geometric Image Processing. SPIE Visual Com-
munications and Image Processing, Lugano, Switzerland.
[47] Yitzhaky Y., Peli E.: A Method for Objective Edge Detection Evaluation and Detector Parameter
Selection. IEEE Transactions on Pattern Analysis and Machine Intelligence, 25(8), 10271033.
[48] Zhang J. K., Davidson T. N., Wong K. M.: Ecient design of orthonormal wavelet bases for signal
representation. IEEE Transactions on Signal Processing.
2004
[49] Lisowska A., Porwik P.: New Extended Wavelet Method of 2D Signal Decomposition Based on Haar
Transform. Mathematics and Computers in Simulation, Elsevier Journal, (to appear).
[50] Porwik P., Lisowska A.: The New Graphic Description of the Haar Wavelet Transform. Lecture
Notes in Computer Science, SpringerVerlag, Berlin, Heidelberg, New York, 3039, 18.
M ach in e GRAP HI CS & VI S I O N vol . 13, n o. 1/2, 2004, p p.79-98 .